In the Spring of 1995, Thomas L. Gravell and the University of Delaware Library agreed to participate with Daniel Mosser and Ernest W. Sullivan, II to make The Thomas L. Gravell Watermark Collection of photographic reproductions of over 7,000 watermarks in paper made between 1400 and 1835 available as a searchable online electronic database on the Internet, with the Experimental Bulletin Board System , hosted by the English department at Virginia Tech, as the Internet server, with Len Hatfield providing technical and design support. The first version (1.0) of the Gravell Archive, using only HTML markup, was mounted on the World Wide Web in 1996 (see Mosser and Sullivan 2000). Relatively quickly, we converted the database to a FileMaker Pro database, served by FileMaker using web pages generated by Claris Dynamic Markup Language (CDML). This version, 2.0, has been plagued with access and reliability problems. In July 2007, Mosser and Radcliffe reconfigured the database so that while data input still employs FileMaker Pro, the database structure is now relational, and employs a series of linked tables. Records from the FileMaker database are uploaded via ODBC to a MySQL database on the CATH server, which generates records on the fly using PHP commands. The current Mac X-Server is located in a secure building, employs RAID backup protection, and is connected by a 100 BaseT data port. We believe version 3.0 will satisfy our needs and those of its users.
This conversion has required some changes in naming conventions. MySQL does not allow fields to contain "NAME" as a searchable entry. Consequently, we have renamed the records that were previously designated as "NAME.XXX.X" as "WORD.XXX.X." The records appear with the desgnation "NamesWords." Other changes have been effected to achieve greater congruency with the International Paper Historian’s Typological Index, and the IPH alpha-numerical designators appear with the Gravell database records, when there is congruency. One major exception is the category "Shield." employed by Gravell but lacking in the IPH Index. The IPH Index refers instead to the content of the shield, which we include under "Secondary Descriptors" as search and descriptive category options. Too many records for "Shield" had already been entered in the database to make their reclassification at this time a practical consideration. We also have a category "Arms" that we use when we can identify a particular heraldic device (e.g., "Arms of London"); this also deviates from the IPH conventions.
In conjunction with the Bernstein Project, we have added a "cf. Les Filigranes" number corresponding to the published watermark number in Briquet and the variant number following the order listed in Les Filigranes. Briquet did not include references to all of the variants he collected, nor even to all of the main types he collected, so in some cases this number cannot be supplied.
The database’s new configuration provides three broad categories for searching: "DESCRIPTOR FIELDS", "WATERMARK FIELDS," and 'ARTIFACT FIELDS." Within these categories, search options are primarily constrained by pull-down menus. Selections within the three categories may be combined: thus, one could select "Lamb" from "DESCRIPTOR FIELDS: ENGLISH LANGUAGE" and "Briquet" from "WATERMARK FIELDS: COLLECTOR" to delimit only the "LAMB" watermarks collected by Briquet that are included in the database.
Thomas Gravell made his original photographic reproductions of the watermarks with the photographic process he developed employing Du Pont DYLUX® 503 photosensitive paper. Far archival purposes, Thomas Gravell rephotographed his original images as slides that are held at the University of Delaware Library, as are the notecards on which Thomas Gravell recorded information identifying the watermarks. The originals of the watermarks are in manuscripts and books at the Folger Shakespeare Library, The Winterthur Library: Joseph Downs Collection of Manuscripts and Printed Ephemera, The Historical Society of Delaware, Hagley Research Library, the Library of Congress, and the University of Delaware Library. The libraries holding the artifacts in which the watermarks occur and the University of Delaware Library have given Daniel Mosser and Ernest W. Sullivan, II permission to scan these slides into digital images and to utilize the information on the notecards for this electronic database. Watermark reproductions and data files derived from these sources will contain a "Gravell Index Number" in the data record; this number corresponds to the number assigned to the watermark by Thomas Gravell. Because the Archive is organized somewhat differently, this number is used only as a secondary reference: e.g. "C5" (the Gravell Index Number) as contrasted with "UNI.001.1" (the Archive reference).
At present, the database is a fraction of its contemplated scope. In addition to the 7,000+ watermark images in the Thomas L. Gravell Collection as well as to add additional photographic watermark images from any source willing to supply such images and identifying information. In 1999, we received permission from the Bibliothèque de Genève (then known as the Bibliothèque publique et universitaire) to add the unpublished tracings collected by Charles-Moïse Briquet (some 27,000). In 2000, we received a "Millenium Grant" from the then-College of Arts & Sciences at Virignia Tech that allowed us to have the unpublished Briquet tracings scanned as TIFF images and burned onto a CD-ROM. An inventory of the Bibliothèque's holdings of Briquet materials, and an annotated English translation are now included in this Archive. Sample entries from photocopies of unpublished Briquet tracings have been added to the archive (to locate these, enter the search term "Briquet" in the search field "WATERMARK FIELDS: COLLECTOR").
The DYLUX® watermark images on slides have been scanned with Nikon CoolScan at a resolution of 109 dots per inch and their contrast sharpened with Adobe Photoshop®. Archival scans (TIFFs at 300dpi) are created at the same time. For images taken directly from DYLUX® paper, we have used a UMAX flatbed scanner. Each image takes up about 100-130 kilobytes in JPEG/GIF format.